- Aim: Linking COVID-19 and demograhic/population data - and have fun with R :)
Hypothesis: That factors like urbanisation and median population age affects COVID-19 kinetics
- Global map - days from 100 cases to 100 deaths
11 5 2020
Aim and hypothesis - inkl søjleplot og map plot Mette
Materiale - oversigt over dataset - Mette Christof
Metode 1 - clean flowchart - Mette Christof
Metode 2 - analyse flowchart - Mette Christof
Covid-19 cases and deaths in each country
Results - variable selection
- Liniar correlation analysis lm(days from dec1 to 100 cases ~ diff. variables)
- Significant correlations:
| Estimate | Std. Error | t value | Pr(>|t|) | |
|---|---|---|---|---|
| (Intercept) | 200.098826 | 11.376201 | 17.589248 | 0.000000 |
| life_expectancy | -0.960407 | 0.192274 | -4.995000 | 0.000002 |
| population_living_in_urban_areas | -0.166332 | 0.062027 | -2.681600 | 0.008278 |
| respiratory_diseases | -156.620453 | 54.571333 | -2.870013 | 0.004793 |
Respiratory diseases
Life expectancy
Population % living in urban areas
PCA analysis by population demographics
- PCA showed a clear association with COVID-19 kinetics
- Relative COVID-19 deaths were more informative than absolute deaths
- PC1 comprises 44.6% of variation
PCA and cluster analysis
- Cluster analysis (n=3) based on population demographics data (middle) and on PCA (right)
- Cluster analysis does not capture COVID-19 kinetics accurately
COVID-19 app
- Confirmed cases, test and death in time series
https://r-kursus.shinyapps.io/covid_app/